Figure 3. Conceptual Research Frameworks Scenario design for
service failure
Pre-testing and modifying Scales
Determine the sample size and analysis method
Execute the sampling process
Data collection
Design scenarios with the same service failure and different recovery speed and different types of explanation.
Choose 10 participants for each scenario and make sure the efficiency of scenarios and scales.
Sample size would be 2 (recovery speed:
immediately/delayed) x 2 (type:
outcome/process) x60 (participants). Let participants are composed of nearly half male and half female.
Randomly assign each participant to a cell.
3.2 Designing scenario
A described scenario was used in this study for several reasons. First of all, Smith et al.
(1999) showed that this scenario method could avoid the problems of intentionally imposing service failures on customers and minimizing memory-bias. This kind of bias was frequent encountered in self-reports of service failures in survey designs. By minimizing memory-bias, researchers could prevent participants from overstating their service failure experience.
Second, as Smith & Bolton (1998) indicated that scenarios could be more effectively manipulated than in real-life settings. The scenario method could create greater variability in customer responses to service recovery than observation in a natural environment could provide. Besides, observing people in actual situation would be both costly and very time consuming. Third, the scenario method reduced problems such as individual differences in responses and personal circumstances to fit into the research context. In other words, the scenario method enhanced internal validity by being able to control extraneous variables and manipulated them (Bateson & Hui, 1992). For all these advantages, many researchers choose the scenario method to avoid- or at least reduce- all of the potential limitations and problems.
The scenario method is most successful when the designed scenario is highly congruent with participants’ experience, so that participants can easily imagine the experimental scenarios (Dabholkar, 1996). Therefore, we ensured that the scenarios were realistic and participants were familiar with the situations described in the scenarios. Although the scenario method is
not without deficiencies, we believed its advantages make it well-suited for this research.
3.3 Sampling plan
A 2 × 2 × 2 between-subject factorial design was used to test our predictions: the influence of recovery speed and toward customer satisfaction, and repurchase intention, with distinct types of explanation and different genders. There were two levels of recovery speed (immediately and delayed), two types of explanation (the focus on the outcome in the service recovery and the focus on the process in the service recovery) and the two genders. There were four kinds of scenarios in this study because the gender of participants was not manipulated. Sixty participants were asked to participate in one scenario independently, which meant the sample numbers would be 240 (60 × 4 = 240).
Participants were exposed to a written scenario describing a service failure in a computer repair station. The scenarios are attached as Appendix I. Participants were told that it was a study about consumer behavior and were given a questionnaire. The questionnaire contained three major parts. The first part described the scenario. Participants were asked to read the scenario carefully and image themselves in it. The second part listed some questions about (CS) customer satisfaction, (RI) repurchase intention, (RS) recovery speed and (ToE) types of explanation. The third part contained demographic information. The questionnaire is attached as Appendix III
3.4 Measurements 3.4.1 Recovery speed
There were two levels of recovery speed in this study, immediate recovery and delayed recovery, with recovery speed being defined as, “how much time did the attendant in the repair station spend on fixing the computer?” An immediate recovery took 15 minutes and a delayed recovery took 3 days. As used by Wirtz & Mattila (2004), two 7-points items with high reliability (r = 0.81) were used in this study. In order to fit the scenario, some words were modified in this study. The two items were, “The attendance was quick in doing some recovery,” and “The length of time taken to give recovery was longer than necessary.”
Seven-points represented “extremely agree” and one point represents “extremely disagree.”
3.4.2 Types of explanation
There were two kinds of explanations in this study, which focused on outcome or on the process. That for the outcome was, “After the component is changed and repaired, it will never have the problem about detecting no wireless network. And it will be more comfortable
and convenient to send and receive mails and to surf the internet. We are sorry to cause the inconvenience and beg your pardon.” On the other hand, the response which focused on process was, “After checking out the in-store data, we found that the component which has to be replaced was out of production. We do not have the component in this station because it
was sold out. We have already back ordered the component from other repair station.
However, it will take longer for repairing your notebook. We are sorry to cause the inconvenience and beg your pardon.” The 7-points Likert scale was chosen, and participates were asked “Do you think the explanation is focus on the process/outcome?” Seven-point represents “strongly agree” and one-point represents “strongly disagree.”
3.4.3 Customer satisfaction
The 7-points Likert scale was chosen here as well, the seven being, “strongly agree” and one point being, “strongly disagree.” Following prior research and modifying the words to fit this study, customer satisfaction after the service recovery was measured using a three-item scale. Three items are “The service provided by the repair station is satisfied”, “Deciding to come to this repair station was not a good decision” and “I am satisfied with the service provided by the repair station.”
3.4.4 Repurchase intention
In Blodgett, Hill, & Tax (1997), customers’ repurchase intention was measured with three items. The resulting scales were highly reliable; Cronbach’s alpha for it was 0.91.Just like the measurement of customer satisfaction, some descriptions of the items were modified for more fit the scenario in repurchase intention. Three items were “The likelihood that I would consume at this repaired station in the future is high”; “If this situation had happened to me I
would never consume at this repaired station again.” and “If this had happened to me I would still consume at this repaired station in the future.”
3.5 Data collection
The data was gathered from 40 participants in the pilot study and 240 participants in the main study. Data were collected via two major channels. In first the questionnaires were handed out to students in the classes at National Chiao Tung University (NCTU) and National Taiwan University (NTU) through. The second was using the Internet to distribute and collect questionnaires online. Four questionnaires with different scenarios were mixed and were given randomly to subjects. All participants were told about the purpose of this study, and were asked to complete the questionnaires carefully. Afterwards all participants were thanked with a small gift.
3.6 Manipulation check
One manipulation check was conducted to test if the recovery speed of the repair station was immediate or delayed. Another manipulation check was conducted to test whether the explanations given by the attendant were focused on the outcome or on the process. The scale items were mentioned in 3.4.1 and 3.4.2 above. The results of the manipulations are reported in chapter four.
3.7 Pretest
A pilot study was conducted to test the reliability of the questionnaire. This is a common method used to discover problems or misunderstandings in the design of the experiment, which can then be modified before the main study. After three failures and subsequent modifications, the fourth trial of the pilot study was successful.
The pretest was made by giving 40 participants the experimental questionnaires, and telling them the research was concerned with consumer behavior. There were 21 male and 19 female participants. Twenty-two of the 40 participants were students.
The reliability of the customer satisfaction scales was 0.759 (Table 1) and the reliability of the repurchase intention scale was 0.906 (Table 2). Both were higher than 0.7, and hence there was a significant difference between immediate and delayed of recovery speed group (p<0.00).
The difference between the groups where explanation focused on outcome or on process was also significant, too (p<0.00).
Table 1. Reliability Statistics of Customer Satisfaction Cronbach's Alpha N of Items
.759 3
Table 2. Reliability Statistics of Repurchase Intention Cronbach's Alpha N of Items
.906 3